Forecasting Control System and Method

This invention concerns a forecastirg control system which takes measured data, and processes it to provide selective data reports including forecasts. The systen makes use of arrays, and in particular arrays embedded within computerised spreadsheets, to process the data and provide the reports. In further aspects the invention concerns a method of operating the computerised spreadsheets and a computer system.

The invention comprises a framework for efficiently generating and controlling forecasts from historical data, and arose in financial forecasting originally. With the advent of personal computers, a number of forecasting packages have been introduced. A typical accountancy package will provide for sequential entry of historical accountancy data, and a computing "engine" will process this data to provide accountancy output, which is compared with a budget. The "engine" will generally be invisible to the user, and its operation need not be understood by the user. In fact, it is commonly the case that the user will not be able to determine how the "engine" is processing the accountancy data to produce the output. Some systems, in an endeavour to reduce the ability of users to tamper with the input data, restrict users ability to change data once entered. These packages tend to be inflexible in their provision of reports, do not usually provide any forecasting facilities, and do not have provision to include technical non-accounting data. Other packages concentrate on forecasting, with minimal or no historical data. Yet other spreadsheet systems are developed to forecast results for companies for economic investment appraisals, but do not include basic accounting controls such as double-entry bookkeeping, applied to forecasts.

The present invention provides a forecasting control system comprising: arrays of cells for the entry of data items representing operational, historical and forecast data for each of a number of defined reporting periods within a time period; wherein each cell in each array corresponds to a particular item of growth or category of assets in respect of a particular reporting period; some of the arrays are designated data entry arrays and data concerning different items of growth are entered into different data entry arrays, each cell of a summary of change factors array is arranged to automatically display data which is the sum of the data entered for the respective item of growth and for the respective reporting period on all the data entry arrays; and each cell of a closing array automatically displays data which is the sum of: the data entered for the respective category of asset and the respective reporting period on all the other arrays; plus the value of the respective cell of the closing array in the immediately preceding reporting period. A note on the terms used is indicated on Annexure A of this application.

The array approach allows the user to define and select in advance the data to be tracked by the system. Both historical data such as particular revenue, expenses, equity, assets and liabilities, and operational data such as measurable technical features of transactions with major groups such as customers, workers, and suppliers, may be entered.

The arrays may be partitioned into sections, each array having a growth factor section and a system status section. Alternatively, separate arrays may be provided for growth factor data, and for system status data, and these arrays may be paired.

Columns in the arrays may record data for reporting periods preselected by the user within a time period, for example the twelve months July to June of a financial year in accounting applications; or the ten years 1990 to 2000 in economic appraisals.

Rows in the arrays may record items of data, for example: sales, purchases, fixed asset totals, creditor totals or other data, and units sold, labour hours worked, orders delivered late or other non-financial data.

Any entry for a particular reporting period may require both a positive value such as a debit for an expense or asset, and a negative value such as a credit for income or a creditor, to be entered in the same column. The action of this control is shown in a check row to clarify that entries in a particular reporting period total zero, or in other words that total debits equal total credits.

The arrangement preserves double-entry bookkeeping, with which bookkeeping staff are familiar, and allows data to be entered by staff without advanced qualifications. The data is easily reviewed from period to period by qualified personnel.

Additional columns may be added to show, for instance, year-to-date actuals, and forecast end-term results.

A summary of change factors (or profit) may may be derived by formulae which automatically sum items affecting profit or chance from data input arrays over the defined reporting periods. The total profit or net change for each month may be recorded in a special profit row in the array.

A summary system status or closing (balance sheet) array may be derived by formulae which automatically sum the status or asset and liability items, plus the opening value of these items at the beginning of each reporting period, plus the total profit derived from the profit array.

Budget data, set at the beginning of the accounting period, actual data recorded during the accounting period and forecast data may be treated in similar ways.

Before a forecasting exercise is started, a budget array is completed showing budget data for each reporting period.

Another array is there, filled with forecast data for each reporting period, showing expected data rather than target or budget data. Actual data is then entered in place of forecast data, as each reporting period is completed. A close relationship of actual and forecast data is maintained. At the beginning of the accounting period, the forecasts may be close to budget, but after a month or two forecasts may differ substantially from budget. This will indicate to management that results for the full year may be substantially above or below original budget, and provides early warning to management to take action to avoid adverse results. By including forecasts, the effect of action by management, such as a decision to raise prices by 10%, can be obtained to determine whether the action proposed will yield the desired result.

The forecasts may be revised in the light of each month's actual results, every month. The forecasts will therefore reflect the most up-to-date information. The forecasts of each item grow naturally out of, and can be seen to be consistent with, the immediately preceding data. Furthermore, the forecast data is subject to the same data controls as actual data. For financial data, a forecast balance sheet is generated for every reporting period in advance, and this forecast balance sheet must balance, as does a balance sheet with historical data. For non-financial data, causal factors are specified and separately forecast, and actual data for the causal factors isolated is reconciled with total system data by an error array.

Another advantage is reversibility. This may be particularly important in analysis of financial statements of public companies. An analyst will not know in advance all the causal factors, that is the accounting transactions which cause the changes in balance sheets. However, from the balance sheets available in summary form, the causal changes may be deduced by disaggregating the data back into individual data arrays for respective items.

The technique of adding and subtracting arrays of data is a general technique that can work in either direction: from the system "balance sheet" to causal factors, or from causal factors to a system "balance sheet".

Reports may be defined by the user. These may be combinations of summary data, comparing budget, actual and forecast data, and comments and graphs as desired. For example, a graph may plot budget levels of cash at bank month by month, and compare these with plots of actual cash at bank each month to date, and forecast cash at bank levels for the remainder of the year. Financial data and non-financial data may be highlighted in the reports as preferred by the user.

The system described can be implemented in a computerised spreadsheet application. In this case each set of revenue and expense and balance sheet data may be placed on a respective page within the spreadsheet. The cells of the sheets may contain formulae linked to the contents of other cells. In particular, some cells in journal arrays may contain formulae which reverse entries made in other revenue or expense cells. In addition, the cells of the closing profit array and the final balance sheet array may contain formulae to add the contents of the respective cells of the other arrays.

In another aspect, the invention provides a forecasting control method comprising the steps of:

displaying a plurality of data entry arrays containing cells for the entry of data representing growth and asset items for each of a number of defined reporting periods within a time period, wherein each cell in each data entry array corresponds to a particular item of growth or category of assets in respect of a particular reporting period;

entering items, historical or forecast, concerning different growth and assets data in the cells of different data entry arrays;

displaying a summary of change factors array in which each cell automatically contains the sum of the data entered for defined items of growth and respective reporting periods on all the data entry arrays; and

displaying a closing array in which each cell automatically contains the sum of: data entered for the defined category of asset and reporting period on all the other sheets plus the value of the respective cell of the closing array in the immediately preceding reporting period.

In a further aspect the invention provides a method of operating a computer spreadsheet application to provide forecasting control comprising the steps of; constructing a plurality of arrays each having the same corresponding elements in which there is identified the same item of growth or category of assets in respect of the same reporting period;

designating one of the arrays as a summary of change factors array and inserting a formula into each growth factor cell to automatically calculate the sum of the entries in the corresponding cell in each of the other arrays;

designating another array as a closing array and inserting a formula into each asset cell to automatically calculate the sum of the entries in the corresponding cell of each of the other arrays plus in the first reporting periods an opening value and in the subsequent reporting periods the value of the respective cell of the closing array in the immediately preceding reporting period.

In a further aspect the invention provides a computer system including a spreadsheet application arranged to provide forecast control comprising a plurality of arrays each having the same corresponding elements in which there is identified the same item of growth or category of assets in respect of the same reporting period;

one of the arrays is a summary of change factors array in which there is a formula in each growth factor cell to automatically calculate the sum of the entries in the corresponding cell in each of the other arrays;

another array is a closing array in which there is a formula in each asset cell to automatically calculate the sum of the entries in the corresponding cell of each of the other arrays plus the value of the respective cell of the closing array in the immediately preceding reporting period.

In a further aspect the invention provides computer software to control the operation of a computer spreadsheet application to perform the method of operating a computer spreadsheet application to provide forecasting control reports, comprising the steps of: constructing a plurality of arrays, each having the same corresponding elements in which there is identified the same item of growth or category of assets in respect of the same reporting period;

designating one of the arrays as a summary of change factors array, and inserting a formula into each growth factor cell to automatically calculate the sum of the entries in the corresponding cell in each of the other arrays;

designating another array as a closing array, and inserting a formula into each asset cell to automatically calculate the sum of the entries in the corresponding cell of each of the other arrays, plus the value of the respective cell of the closing array in the immediately preceding reporting period.

Examples of the invention will now be described with reference to the accompanying drawings.

FIG. 1 is a schematic diagram outlining a prepared embodiment of the invention.

Table 1 presents known Australian population at end of various specified periods.

Tables 2 to 9 concern a first example showing the application of the invention to records of the Australian population;

Table 2 is an array of population births;

Table 3 is an array of migrations to Australia;

Table 4 is an array of population deaths;

Table 5 is an array of permanent departures,

Table 6 is an array of net interstate migration;

Table 7 is an array of statistical errors;

Table 8 is an array of the sum of change factors; and

Table 9 is an array showing population at the close of each year.

Table 9a shows the total population at year end.

Table 9b presents regression forecasts for specified years.

Tables 10 to 13 concern a second example applied in an accounting scenario.

Table 10 is an array of cash receipts;

Table 11 is an array of journal entries;

Table 12 is an profit and loss statement; and

Table 13 is an balance sheet.

Referring first to FIG. 1 the reporting system 1 has a structure comprising two identically structured sub-systems 2 and 3 feeding information to the reports 4. The first sub-system 2 concerns budget data, and the second subsystem 3 concerns actual and forecast data.

In both sub-systems 2 and 3 there are data entry arrays 5 and 6. Economic data is entered into arrays 5 and operational data is entered into arrays 6. Change factor data and system status data both go into arrays 5. There are multiple arrays 5 each having the same dimensions, and concerning the same reporting periods and data. Data entry is disaggregated by entering different economic data items into different sections of the arrays 5. Arrays 5 are summed to produce a summary of change factors in array 7, and a final asset status summary in array 8. Operational data is shown as feeding direct to reports 4; however operational data may also be split into change factor data and system status data in the same way as the economic data. Arrays 6, 7 and 8 feed to reports 4.

In accounting applications, financial and non-financial data is input to arrays 5 and 6 respectively. The non-financial data input array 6 feeds direct to the reports 4. The financial data input arrays 5 are summed to produce a summary array 7 of profit, and another summary array 8 of closing balance sheets.

In both cases, the principles of double entry bookkeeping are applied to ensure an important control over the data, for each reporting period:

It is not difficult using formulae to generate forecasts; application of this control however requires the forecasts to be generated in such a manner that the system is balanced. This enhances the reliability, consistency and verisimilitude of the forecasts. If asset acquisitions are forecast or budgeted, for example, the system requires liabilities (or diminution in cash reserves) to be also forecast or budgeted in exactly equal amounts, or error messages will be generated.

In a first example the population of Australia between 1986 and 1991 will be recorded and processed in order to provide forecasts for the population through 1992 to 1997. The base data is given in Table 1.

1992:0

PO 150

Tables 2 to 9 are set out showing data for the years 1986 to 1997 in columns. The first six rows show growth items. The following eight rows show a breakdown according to states and territories. The ninth row shows a control total which should sum the column to zero, and a final row present on some sheets shows a rate based on the population at the end of the preceding year.

Table 2 deals with population births during the year. In the first row the actual birth figures are inserted for the years 1987 to 1991 and the breakdown of these figures by states and territories is set out lower down in the same columns. Forecasts are then made for the years 1992 to 1997 on the basis of the birth rate established from the historical data.

Table 3 deals with migration to Australia and again the actual figures are entered from 1987 to 1991. For 1992 to 1997 the forecast is for migration at the 1991 rate.

Table 4 deals with deaths. The actual figures are placed for 1987 to 1991 and the forecast figures are generated using the death rate of the historical figures.

Table 5 deals with permanent departures. Historical data is entered from 1987 to 1991 and forecasts are made at a constant rate of departure for 1992 to 1997.

Table 6 shows the net interstate migration and although this has not been forecasted the overall effect on the country's population is effectively zero for all years.

Table 7 shows the statistical errors in the historical data.

Table 8 is a summary of the change factors. This summary shows the growth items taken from Tables 2 to 7 and displayed in an array in corresponding location to those occupied in the earlier figures. It can be considered as a consolidation of Tables 2 to 7 and may be derived by adding the corresponding growth items of Tables 2 to 7 together and displaying them. An additional line is entered beneath the table of growth items showing the net change throughout the year.

Table 9 is a snap shot of the total population at the end of each year. It shows the actual and forecast breakdown of the population into the various states and territories. This array also shows the 1986 values and the net change row from Table 8. The previous year's population can be added to the total population at the end of the preceding year shown in the row labelled "Previous Investment" in order to arrive at a total population figure for the year end. The total population figure for the year end is not shown in Table 9 but is shown in Table 9a.

Table 9a also shows the actual and forecast population figures from the 1996 year book and this data has been used to test the forecast populations and illustrates these by total error and percentage error.

An alternative way of forecasting the data is to apply a simple regression to the total population Table as shown in Table 9B. The errors shown using this aggregated approach indicate higher errors than the forecasting which embodies the present invention. This does not say that the matrix disaggregation approach of the present invention will always yield better results than regression formulae.

A second example will now be described with reference to Tables 10 to 13.

The arrangement of the data in arrays and in particular the separation of the data into growth factors and asset categories is compliant with standard double counting bookkeeping practice. This makes the invention particularly suited to accountancy applications.

Table 10 shows an array of twelve columns each headed by the name of a month from July to June. The rows show growth factors as profit and loss entries divided into revenue and expenses and the system status categories as balance sheet entries divided into assets liabilities and equity. There is a totals row at the bottom of the page.

Data is credited (negative figures in brackets indicate a credit) into the row representing sales in the profit and loss array and the corresponding debit entry is made in the cash at bank row on the balance sheet array for cash sales. The total row at the bottom shows zero entries as a check.

The first two columns July and August are sub-headed actual and the data in these columns is actual data from subsidiary records maintained by bookkeepers. The remaining columns are sub-headed forecast and include data which is estimated.

Table 11 shows an array which is identical in layout to Table 10 but which has different entries. The entries on this array are journal entries and represent depreciation. Depreciaton is debited as an expense in the profit section and credited representing an addition to the provision or diminution in value of the fixed assets in the balance sheet section. Again the entries for each month total to zero.

Table 12 again shows identical arrays to Tables 10 and 11 but the data shown in the cells of the profit and loss array is the sum of all the data shown in the respective cells of the profit and loss arrays of Tables 10 and 11. Entries are automatically generated in the profit row equal to the sum of the other rows.

Table 13 is again an array of identical layout but data is only shown in this case in the balance sheet section. An additional column is provided on the left hand side of the array for the opening values. The values in the first column of the array that is July represent the sums of the corresponding values in all the preceding sheets added to the opening values. The sums shown in the second column August are the sums from the corresponding locations in all the preceding sheets but in addition they are added to the preceding month values (July). The process is continued across the array. The profits shown in the profit row in the profit and loss statement Table 12 are also input onto the balance sheet.

An identical set of arrays could be provided for budget data.

Realisation

In a practical example a business entity finds through experience that certain journals are required to record accounting transactions: a sales journal, a purchases journal, a cash receipts journal, a cash payments journal, a general journal, for example. Let n=the number of journals required for a particular business entity.

Then consider a matrix A of 3 dimensions; this may be visualised as a spreadsheet consisting of (n+2) sheets. The first n sheets contain the journals required by the business entity.

Sheet (n+1) contains an array of profit data.

Sheet (n+2) contains an array of balance sheet data.

In general terms let A(s i j) denote an element of matrix A corresponding to sheet s account i period j. For example in all entity with 5 basic journals (n=5) the matrix A or spreadsheet can take the following form:

s=1: Sales journal data

s=2: Purchases journal data

s=3: Cash receipts journal data

s=4: Cash payments journal data

s=5: General journal data

s=6: Profit statement

s=7: Balance sheet

where i denotes an account number in the entity's chart of accounts and j denotes a month number in the current accounting period (financial year) of 1 2 3 . . . 12.

The journal matrices may be partitioned matrices containing profit data and balance sheet data in separate sections.

The profit statement on sheet (n+1) is derived by adding the profit data in the first n sheets for any particular month j: ##EQU1## for i=revenue or expense account.

The profit is found by adding all entries i for that month; in the profit statement and recording the sum in a special row p: ##EQU2## p is conveniently located as a row in the equity section.

The balance sheet on sheet (n+2) is derived by adding the balance sheet data in the first n sheets for any particular month j together with the corresponding balances of the previous month: ##EQU3## for balance sheet accounts i.noteq.p

The invention also postulates that the profit for the current month found in the profit statement is added in to the equity in the balance sheet statement:

If all accounting entries are correct then the data in the balance sheet in each month j will sum to zero: ##EQU4## for j=1 2 . . . 12.

Whatever the current month say j=t. This means that both actual and forecast data are subject to the same overall control: total debits are balanced with total credits whether they are historical debits and credits or future debits and credits. The advantage of this feature in forecasting is that it is a check to ensure that liabilities are not omitted. Only by including all forecast liabilities will a forecast balance sheet balance.

It is important to ensure that in the arrays s=1 2 . . . n recording journal data input that the sum of all entries to all accounts in any month j will add to zero. This summation to zero is obtained on the journal input matrices by entering the complete accounting entry for each entry each month; that is ensuring total debits equal total credits. This is a useful control measure in its own right but also ensures matrices can be added together to yield a profit statement and a balance sheet.

Full year results are anticipated by inclusion of forecasts in each data cell for each account each month up to the end of the year on the same double entry accounting principle for actual results.

forcast Profit to year-end ##EQU5## is founded by summing actual profits for months 1 2 . . . t plus forecast profits for months (t+1) (t+2) . . . 12.

Forecast balance sheet data is obtained in a similar way by summing actual data in month t plus forecast data for months (t+1) (t+2) . . . 12.

Forecast data is therefore subject to the same data control as actual data.

This logic is equally applicable to budget data:

A budget profit statement is obtained at the beginning of a year by adding budget journal matrices for revenue/expense accounts.

A budget balance sheet is obtained at the beginning of a year by adding budget journal matrices for asset/liability/equity accounts.

A report to management is obtained by comparing key budget and actual/forecast data:

Current month: Actual results and budget results

Year to date: Actual results and budget results

End Year: Actual results for the first t months plus forecast results for the remaining (12-t) months for the full year compared to full year budget results.

A slight variation entails set-up of a matrix to hold opening balances on accounts at the beginning of each month; the profit statement could be held on sheet n+2 and the balance sheet held on sheet n+3.

Then A(1 i j)=A(n+3 i j-2) for j=2 3 . . . 12 and A(1 i 1) has data input from the previous financial year.

Other variations may involve for example additional cash receipts and cash payments sheets/matrices for additional bank accounts.

Also all matrices may be partitioned and the profit statement and balance sheet combined on one sheet which could be called a trial balance matrix.

Another application is to annual results comparing annual actuals with a long term plan for a defined period of t years as budget:

For example a planning cycle or term of t=5 years;

In general:

Each year within the planning term of t years:

Compare actual results and budget results for each year

But also: Compare (a) the sum of actual results for the first j years and forecast results for the remaining (t-j) years of the planning term or cycle; with (b) the planned or budget results for all t years of the planning term or cycle

Even though the invention has been described with reference to particular embodiments it should be appreciated that it may be embodied in many other ways. For instance it is not necessary for the arrays to be identical nor for them to be positioned in the same places on respective sheets since it is possible to add elements from different arrays together regardless of their positions. However it is convenient and advantageous to the user to employ that arrangement. Error messages may also be provided to highlight any errors in data input or forecasts, in order to flag new trends or input errors. The source of all data changes could also be identified and recorded.

Although the invention has been described with reference to applications which are not technical in themselves, it is possible to apply the invention technically. The invention could be employed in the many different control systems such as:

FORECASTING OF MINERAL RESOURCES

Objective: Forecast specified mineral reserves, Australia 1997-2005

System status categories: Estimated reserves at specified locations

System change categorles: Production, destruction, evaporation, wastage

Unit of measurement: Kilotonnes or megaliters

FORECASTING OF BIOLOGICAL RESOURCES

Objective: Forecast (eg) Australian forest resources, 1997-2005

System status categories: Estimated reserves of different species of trees on Australian commercial plantations

System change categories: Sawmilling, destruction by fire, planting, additions purchased, relinquishments

Unit of measurement: Hectares

FORECASTING OF MILITARY RESOURCES

Objective: Forecast (eg) Australian army personnel, 1997-2000

System status categories: Numbers of personnel of different ranks and different training at different locations

System change categories: Recruitments, killed, resignations, promotions, transfers

Unit of measurement: Number of personnel

FORECASTING OF A CITY OFFICE SPACE

Objective: Forecast (eg) stock of office space, Melbourne 1997-2000

System status categories: Office space in 6 city areas defined by the Building Owners & Managers Association

System change categories: Supply additions, vacant, withdrawals

Unit of measurement: Square meters

FORECASTING OF VEHICLE FUEL USAGE

Objective: Forecast (eg) fuel consumption for a satellite launch rocket every 1 second after ignition

System status categories: Quantity of fuel remaining in different fuel cells during satellite launch

System change categories: Fuel flow for production, wastage

Unit of measurement: Liters

FORECASTING OF SMALL BUSINESS

Objective: Forecast number of small business enterprises Australia 1997-2000

System status categories: Numbers of small firms by location and/or year of start-up and/or type of firm

System change categories: Startups, Failures by cause of failure

Unit of measurement: Numbers of firms

In these environments the ability to forecast disaggregated items, such as the behaviour of fuel cells individually, could be used to ensure that the behaviour of the rocket remained within operational norms. Circumstances where actual behaviour differ substantially from forecast behaviour could indicate failures, such as fuel leaks or engine damage, and could even trigger a decision to abort if the measured data indicated too great a variance from the forecast data.

The invention may find application in many controlled environments besides the particular financial environments described.

It will be appreciated by persons skilled in the art that numerous variations and/or modifications may be made to the invention as shown in the specific embodiments without departing from the spirit or scope of the invention as broadly described. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive.